Parametric study of a VLSI plastic package using locally refined finite element models

Abstract

VLSI package designs continue to utilize relatively large dies in smaller packages. Elevated thermal stresses are likely to occur in such configurations during fabrication, testing, and operation of these packages. This paper presents a parametric study of thermal stresses in plastic packages that are induced during the die attach and encapsulation fabrication steps. Finite element models (FEMs) of plastic VLSI packages were used to vary package design parameters (die thickness, die bond materials and thickness, lead frame thickness, and package thickness and width/length) over typical ranges. Areas of complexity were considered in more detail using locat finite element analysis. The loss of adhesion between the mold and the die and/or lead frame in the package was simulated using gap finite elements in order to consider the extreme cases of no slip or no adhesion at these material interfaces. Changes in die thickness, die bond thickness, and package thickness caused significant changes in package thermal stresses; whereas changes in lead frame thickness, package width/length and die bond elastic modulus corresponded to minor changes in package stresses. The loss of adhesion at the mold material interface resulted in significantly altered thermal stress fields, e.g., the stresses were concentrated near the corners of the chip and lead frame where cracking has been observed in DIP plastic molding materials.

Thermomechanical Finite Element Analysis of Problems in Electronic Packaging Using the Disturbed State Concept: Part 1—Theory and Formulation

“…The micromechanically based model developed by Zubelewicz et al (1990) has been proposed for crystalline materials such as the Pb96.5/Sn3.5 solder. Subrahmanyan et al (1989) proposed a damage integral approach for the thermal fatigue of solder interfaces. Another micromechanical constitutive model was proposed by Kashyap and Murty (1981) based on tests the authors conducted on bulk specimens of eutectic solder alloy.…”

“…Finally, Lau et al (1987) proposed a material model based on the yon Mises yield surface, but the model does not include time dependent creep or relaxation features. Most of the finite element procedures use yon Mises or Tresca type classical plasticity material models, Lau et al (1987) and Simon et al (1989). Coffin-Manson curves are commonly used to predict thermal fatigue lifetime, Oshida and Chen (1991), Tien et al (1991), Dasgupta et al (1992), Lau et al (1987), Tummala and Rymaszeski (1989) and others.…”

Thermal and mechanical finite element analysis of a VLSI package including spatially varying thermal contact resistance

“…When such delamination occurs, the molding and operating stress distributions are significantly different from the stress fields associated with perfect adhesion at material interfaces [6]. In particular, the normal stress (pressure) at a delaminated interface is altered and may exhibit elevated levels at the end of the heat spreader and reduced levels near the center of the package.…”

Accurate prediction of PQFP warpage

“…And deviations from the above assumptions have almost exclusively been concerned with delaminations or loss of adhesion at material interfaces by Nishumar (71, Simon [9], and Kelly [ll]. No consideration (in literature) has been given to looking at the assumption that the package is stress free and undeformetl at the molding temperature.…”